Drugs and Drug Delivery for Acute Myeloid Leukemia

Standard chemotherapeutic regimens for AML treatment are based on a combination of an anthracycline and cytarabine.

6.3.1.1 Anthracyclines

Anthracycline development began in the 1960s ⁵⁶. Most of these agents have to be administered intravenously, except for idarubicin for which an oral formulation is available. Anthracyclines are taken up by the target cell via passive diffusion and, once inside the nucleus, intercalate with DNA. Furthermore, they inhibit strand re-ligation by topoisomerase II, causing DNA double-strand breaks ⁴⁶. After hepatic metabolization, anthracyclines are eliminated by biliary excretion. Daunorubicin is the anthracycline most often used for AML treatment. Its lipophilic analogue idarubicin and its active metabolite 13-hydroxyidarubicin have a longer half-life than daunorubicin. Despite preclinical evidence suggesting otherwise, clinical trials have failed to prove a substantial advantage of idarubicin over daunorubicin in terms of efficacy and toxicity ²⁶⁷. Mitoxantrone is a synthetic anthracycline analogue used in combination with cytarabine for AML with at least comparable, maybe superior efficacy in upfront and re-induction regimens ^{8, 37}.

6.3.1.2 Cytarabine

Cytarabine was approved by the FDA almost 40 years ago. The drug is administered parenterally, for induction regimens usually intravenously, and has a short half-life requiring high-dosed short time or medium-dosed continuous infusions ^{197, 222}. Inside the cell, the phosphorylated drug enters the nucleus and is incorporated into DNA in place of cytosine, blocking DNA replication. Cytarabine is metabolized by cytidine deaminases and is eliminated by renal clearance. Like other chemotherapeutics, its action is cell cycle-dependent, and therefore its therapeutic effects are focused on rapidly dividing cells like cancer cells despite its unspecific biodistribution.

6.3.1.3 Standard treatment for patients in good physical condition

The most common chemotherapy regimen to induce remission in AML is daunorubicin as a 15 minute intravenous injection daily for three days plus cytarabine given by continuous intravenous infusion for seven days (so-called "3 + 7" regimen).

With this regimen, 60-80% of patients, depending on age and other risk factors, achieve a complete remission ^{27, 67}. This response rate has not been improved to a clinically relevant extent by changing the dose of any of the two agents or by adding an additional drug. The cytostatic agents used for remission induction confer substantial toxicity including myelosuppression, mucositis, diarrhea, and cardiotoxicity.

6.3.2 Novel Therapeutic Agents

In view of the high remission rates achieved in AML patients using the standard chemotherapeutic regimens, novel agents would have to meet high standards of efficacy to replace these regimens ⁴⁴. However, relapse rates and toxicity as well as the limited treatment options in elderly patients highlight the urgent need for novel agents that improve disease-free survival and do not add substantial toxicity. While conventional chemotherapy may remain the backbone of treatment, novel agents could be added to improve outcome. Within the last years, many such novel agents have been introduced. Some of them have started to gain the status of a standard treatment option in certain settings, such as liposomal or antibody-conjugated chemotherapy. Others are currently at a more experimental stage, including farnesyltransferase inhibitors ¹¹⁷, histone deacetylase inhibitors ¹³⁶, proteasome

inhibitors ²⁹⁷, or antiangiogenic agents such as bevacizumab ¹¹⁶. Yet, many challenges remain, which are addressed at the end of this article.

6.3.2.1 Liposomal Delivery of Chemotherapeutic Drugs

Anthracyclines are one of the two standard chemotherapeutic drugs in AML.

However, their toxicity is of concern. Above all, cardiotoxicity is dose-limiting and cumulative dose-dependent, which often prevents anthracycline re-treatment in relapsed AML or even upfront treatment in patients with cardiac disease.

To increase the therapeutic index, liposomal formulations have been proposed as carriers for cancer therapeutics several decades ago ⁸⁵. Liposomes encapsulate an aquaeous solution containing the drug inside a hydrophobic membrane. Liposomal encapsulation results in reduced anthracycline uptake by normal, non-neoplastic tissues. In contrast, delivery to tumor tissue and to the bone marrow is enhanced due to the passage of liposomes through fenestrations of the vascular endothelium which are characteristic for these but not other tissues ^{77, 192}. Liposomes are believed to be taken up by membrane fusion rather than endocytosis unless they are modified specifically to trigger this event²⁵². Liposomal formulations are characterized by slower pharmacokinetics compared to non-encapsulated administration of a given drug. They may therefore be the agents of choice when the objective is to maintain a defined plasma concentration with little change over time, rather than high, but quickly decaying, peak levels.

Liposomal formulations of doxorubicin and daunorubicin are currently available for clinical use. The application of liposomal daunorubicin in AML has been extensively reviewed elsewhere ⁶⁸. Briefly, compared to conventional daunorubicin application, liposomal daunorubicin results in reduced conversion into its toxic metabolite daunorubicinol and reduction in toxic side effects such as cardiotoxicity, alopecia, nausea, or myelosuppression. In addition, various in vitro studies suggest that liposomes may help to overcome P-glycoprotein-mediated efflux of anthracyclines, a mechanism believed to contribute substantially to anthracycline resistance in AML and other tumor cells ^{166, 250}. Liposomal daunorubicin combined with cytarabine or alone yielded a complete remission rate of approximately 30% - 45% in patients with refractory or recurrent AML ^{53, 69}.

Liposomes can be targeted by incorporation of homing molecules into their hydrophobic surface. For instance, attachment of folate molecules to liposomes ³⁰¹

via a PEG anchor was used to target cells expressing the folate receptor, a common property of malignant cells in general ¹⁴⁶, and of AML cells in particular ^{156, 214}. The efficiency of such targeting approaches could possibly be increased if the expression of a receptor of interest can be stimulated such as it is possible with all-trans retinoic acid that induces an upregulation of the folate receptor in AML cells in vitro ²⁶⁹.

Efficient liposomal delivery may require sophisticated strategies depending on the drug of interest. For arsenic trioxide, a procedure for the formation of nickel (II) arsenite complexes in liposomes that release the active drug under acidic pH conditions as present in lysosomes has recently been suggested ⁴⁹. Increasing particle stability is an important issue in improving liposomal therapy, but it may be achieved at the cost of impaired drug release. A recently described approach using lipase may overcome this problem ⁵¹.

6.3.2.2 Novel Drugs Interacting with Intracellular Targets

The tremendous success of the BCR-ABL tyrosine kinase inhibitor imatinib mesylate in chronic myeloid leukemia has stimulated the exploration of novel agents targeting various pathways in cancer. For AML, our increasing knowledge about intracellular signaling cascades involved in this disease has revealed a number of promising targets for inhibitory therapy by small molecules. They are usually applied orally and do not depend on receptors for cellular uptake.

One therapeutic approach is directed towards the RAS protein, which is frequently mutated and therefore dysregulated in AML and other malignancies ¹⁸⁰. Attachment of RAS and other regulatory molecules to the plasma membrane is crucial for their functionality. Small molecule farnesyl transferase inhibitors such as tipifarnib and lonafarnib ¹¹², after passively diffusing into the cell, inhibit RAS membrane anchoring.

Tipifarnib has achieved clinical responses in patients with refractory and relapsed poor-risk AML ¹¹⁷ and is currently being evaluated in phase III trials ^{9, 240}.

Another novel therapeutic approach targets the FMS-like tyrosine kinase 3 (FLT3).

Mutations in the FLT3 gene producing internal transmembrane duplications (FLT3/ITD) are common in AML and result in constitutive FLT3 activation ^{138, 179}. A number of small molecule inhibitors of FLT3 have been evaluated in clinical trials lately, including tandutinib (MLN518), lestaurtinib (CEP-701) ²³⁵, and PKC412, and evidence of antileukemic activity has been seen 130, 235, 241. Like other kinase

inhibitors, these agents are orally applicable and their delivery to AML cells is receptor-independent.

While the oral application of small inhibitory molecules simplifies their use in an outpatient setting, this may not always be the preferred way of administration given the poor oral intake and nausea experienced by many cancer patients under treatment ³⁰⁰. In addition, target specificity remains an issue in kinase inhibitor therapy. Under some conditions, inhibitors with multiple targets may have beneficial effects, as shown recently for the multi-kinase inhibitor sorafenib in a xenograft model of FLT-driven leukemia ¹³. Yet, the lack of specificity of some kinase inhibitors may account for limited anti-leukemic activity and side effects. The latter are usually considered mild compared to those associated with conventional cytostatic drugs, but can occasionally be quite severe, e.g. in heart tissue, as described for imatinib and other agents ⁷⁶.

In terms of specificity, agents such as monoclonal antibodies or peptides targeting cell surface molecules may therefore be superior to small molecules.

6.3.3 Receptor-targeted Drug Delivery in AML

Targeting cell surface molecules in cancer is a paramount issue in drug delivery both affecting efficacy and specificity (and therefore toxicity) of an antineoplastic drug. By specific homing after systemic administration, compounds are directed to the cell type or tissue of interest. This prevents their action in non-target tissues, thereby increasing therapeutic efficiency while decreasing adverse effects. Thus, as for other malignancies, drug-conjugated ligands targeting unique surface receptors have been developed for AML treatment.

6.3.3.1 Anti-CD33 monoclonal antibodies

During the last decade, targeted monoclonal antibodies have revolutionized cancer therapy. In AML, the CD33 antigen is a promising target since it is ubiquitously expressed on myeloid blasts in most patients, but neither on healthy pluripotent hematopoietic stem cells nor most non-hematopoietic cell types. CD33 is a member of the sialic-acid binding Ig-like lectin (Siglec) family and has two cytoplasmic immunoreceptor tyrosine-based inhibitory motifs (ITIMs). CD33 is involved in cell-cell interactions and signaling in the hematopoietic system and may have regulatory

functions in the immune system and in cell proliferation ^{143, 188}. The first targeted compound successfully used in AML treatment was Gemtuzumab ozogamicin (GO), a monoclonal anti-CD33 antibody linked to the cytotoxic agent calicheamicin. The conjugate is usually given as a two-hour intravenous infusion. Following systemic administration, GO is efficiently and specifically directed to CD33-positive cells. Upon binding to CD33, the GO-CD33 complex is rapidly internalized. The uptake is boosted by new CD33 molecules replacing the internalized ones ²⁵⁸. Lysosomal release of calicheamicin and translocation to the nucleus cause DNA double-strand breaks and cell death. The efficacy of the drug is influenced both by CD33 expression level and P-glycoprotein activity ²⁶⁸. Consequently, therapeutic efficacy of GO may be potentiated by in vivo stimulation of CD33 surface expression on AML blasts in patients with G-CSF ¹⁴⁷, or by reducing the calicheamicin efflux of malignant cells by P-glycoprotein inhibitors ¹⁷⁸.

GO treatment in patients with relapsed AML can result in remission rates as high as almost 30% 45, 188, 232, 240, 248, 281. As CD33 is also expressed by benign myeloid precursor cells, Kupffer and sinusoidal liver cells, myelosuppression and hepatotoxicity are common GO-side effects ¹⁸⁸. In addition, anaphylactic reactions and veno-occlusive disease have been described as life-threatening side effects in a low but significant number of patients. Other toxicities of GO include fever, hypotension, and abnormal liver function tests, all of which are usually transient ²³⁹. Anti-CD33 antibodies have shown effects against leukemic cells in vitro even without the attachment of a cytotoxic drug ²⁶⁶. However, the unconjugated humanized anti-CD33 monoclonal antibody lintuzumab failed to elicit anti-leukemic effects when added to conventional chemotherapy in a phase III trial ⁷⁰. Nevertheless, the promising studies using GO reveal the potential of targeted drug delivery in AML treatment.

Since FMS-like tyrosine kinase 3 (FLT3) is expressed on approximately 90% of AML cells and plays a major role in survival and proliferation signaling in leukemia blasts, several FLT3 small inhibitor molecules have been demonstrated to show anti-leukemic activity, as outlined above. Nevertheless, the lack of specificity of these kinase inhibitors remains a significant problem as they also interact with several other cellular kinases ²⁴⁸. Furthermore, cellular targets of most chemotherapeutic agents are located in the nucleus, therefore rapid internalization of drug-ligand conjugates is critical to maximize therapeutic efficacy while minimizing side effects. Towards this

end, several FLT3-directed antibodies were isolated using a cell-based phage library screening protocol and two fully human antibodies with the capability to trigger efficient receptor internalization upon binding to FLT3 were generated ²⁸⁰. Such anti-FLT3 antibodies may be promising therapeutic agents in anti-FLT3-expressing AML for receptor blocking or for antibody-guided cytotoxic drug therapy.

For further development of receptor-targeted cancer therapy, a comprehensive understanding of differential receptor expression is needed. So far, very little is known about receptors specifically expressed on AML cells and their interaction during disease development and progression. Some knowledge about unique receptor profiles of AML cells may be gained from microarray gene expression profiling ^{38, 257}. Among the limitations of such approaches is the fact that the protein expression patterns do not necessarily correlate with the functional state and extracellular accessibility of the potential target molecule. Protein-based techniques may be of advantage here, as discussed in the following section.

6.3.3.2 Novel Cell Surface Markers as Potential Therapeutic Targets in AML Phage display is a powerful tool to select for novel ligands targeting cell-type specific surface molecules even if only the cell type of interest rather than an exact target receptor is known a priori. The receptors bound by such ligands can be subsequently identified in the majority of cases. Screening phage displayed human antibody libraries on primary AML blasts, Bakker et al. enriched a single chain Fv fragment strongly binding to myeloid cells. The antigen was identified to be the transmembrane glycoprotein C-type lectin-like molecule 1 (CLL-1). CLL-1 acts as a signaling receptor and is expressed in >90% of AML samples. CLL-1 appears to be restricted to hematopoietic, particularly myeloid, cells. It is also weakly expressed in CD34+/CD38+ or CD34+/CD33+ progenitor cells. Of note, CCL-1 expression is absent in the CD34+/CD38- or CD34+/CD33- stem cell compartment ¹⁵ but may be found in CD34+/CD38- leukemic stem cells ²⁵⁹. Almost 70% of CD33-negative AMLs expressed CLL-1, indicating that CLL-1 complements CD33 as a therapeutic cell surface target for AML. Anti-CLL-1 antibodies may therefore have great potential for AML therapy and for the detection of AML stem cells. This may improve efficacy of current therapeutics, especially when combined with CD33-directed therapy ¹⁵.

A non-biased approach to the identification of high-affinity binding ligands is the screening of phage libraries displaying small random peptides. This strategy has

been successful for a variety of cell types and tissues in vitro and in vivo ^{135, 254}. Linked to cytotoxic agents, such peptide ligands can be exploited for targeting cytotoxic drugs or other therapeutic agents to the cell type of interest 4, 7, 63, 134, 255. Furthermore, screening phage peptide libraries allows for the exploration of epitopes recognized by known antibodies or even the identification of novel molecular markers by fingerprinting of circulating antibodies in cancer patients 25, 26, 167, 264.

In a recent study, we selected phage libraries on AML cell lines. We identified a peptide with the amino acid sequence CPLDIDFYC which strongly and specifically binds to AML cells ¹⁰⁹. Binding correlated with the expression of the AML1/ETO fusion gene which is a result of the the chromosomal translocation t(8;21), the most frequent karyotype aberration in AML. We identified VLA-4 (α4ß1) integrin as a potential receptor for the leukemia cell-binding CPLDIDFYC peptide ¹⁰⁹. VLA-4 is involved in cell-cell and cell-extracellular matrix adhesion by interaction with the vascular cell adhesion molecule VCAM-1 and the extracellular matrix protein fibronectin. Attachment to fibronectin within the bone marrow stroma appears to mediate resistance to chemotherapeutic drugs in leukemia cells ¹⁶⁰. CPLDIDFYC and other VLA-4 antagonists such as the monoclonal anti-VLA-4 antibody natalizumab may therefore serve as future therapeutic agents in AML for receptor blocking or for cytotoxic drug delivery.

6.3.3.3 Leukemic Stem Cells as Potential Therapeutic Targets in AML

Acute leukemia most likely develops from a single transformed hematopoietic progenitor cell. A substantial amount of evidence suggests that, once this cancer has evolved, a subpopulation of leukemia cells with the stem-cell-like characteristics of asymmetric division and self-renewal capacity drives the course of the disease. The characterization of these leukemic stem cells (LSCs) has therefore gained tremendous interest during the last decade. LSCs may withstand cytotoxic chemotherapy as they are often in a quiescent state, unlike their rapidly proliferating progeny ¹⁰⁴. LSCs are therefore considered to be responsible for recurrence of leukemia even after initial treatment success. LSCs have been characterized by the presence or the absence of various sets of surface markers, but are widely recognized to be part of the CD34+/CD38- cell compartment ^{34, 144}.

LSCs may be distinguished from non-malignant hematopoietic cells by the presence of the interleukin-3 receptor α chain (CD123) ¹¹³. This finding has made CD123 a

potential therapeutic target. A diphtheria toxin-interleukin-3 fusion protein has shown toxicity against LCSs while sparing normal progenitors in vitro ^{72, 102}, and such treatment prolonged survival in a mouse model ²⁸. The compound was recently evaluated in a phase I study ⁷⁸.

While markers exclusively expressed on LSCs appear particularly attractive for the purpose of targeting LSCs, there is evidence that certain receptors can be promising therapeutic targets even if they are expressed on other cell types as well. The adhesion molecule CD44 – although expressed ubiquitously – is thought to be crucial to the malignant properties of AML LSCs, and an activating anti-CD44 antibody reduced engraftment of AML cells in a mouse model ¹¹¹.

6.3.3.4 Gene Delivery

Despite many hurdles, gene therapy might be a future option for AML treatment. The spectrum of therapeutic transgenes mediating killing of malignant cells comprises genes encoding toxic, pro-apoptotic, antiproliferative proteins or classical suicide genes such as the herpes simplex virus thymidine kinase gene. Alternatively, immune system-mediated cancer cell elimination may be achieved by delivery of genes encoding costimulatory molecules, e.g. interleukin-2 (IL-2), IL-7, IL-12 ^{62 73, 223}, or immunomodulatory molecules such as CD40, CD80 119, 184, 243, or interferon β ³⁵. One of the major unsolved issues in gene therapy is vector application and delivery to the cells or tissue of interest. Development of efficient and specific vectors for gene transfer is just as crucial to therapeutic success as is the choice of the transgene itself. Currently, viral vectors remain the most effective means for therapeutic gene delivery, although substantial progress in non-viral transduction of hematopoietic cells has been achieved, including electroporation, nucleofection, and particle bombardement techniques ²²¹. Initial in vitro experiments have suggested lentiviral ²⁴³, retroviral, or adenoviral vectors as suitable delivery vehicles for leukemic cells ²¹¹. However, unintended integration of retroviral vectors into the genome or adverse immune reactions elicited by adenovirus administration are serious safety issues to be considered in choosing vectors for clinical application. Over the past years, vectors derived from adeno-associated virus (AAV) have emerged as efficient tools to achieve long-term gene expression in a wide range of cell types. The low frequency of random integration into the genome ⁴⁷, as well as the absence of a

substantial cellular immune response make AAV vectors promising tools in terms of biological safety ^{54 24}.

Various approaches have been taken to make the binding of therapeutic vectors to target cells more efficient and specific. Bispecific conjugates such as antibodies that

Various approaches have been taken to make the binding of therapeutic vectors to target cells more efficient and specific. Bispecific conjugates such as antibodies that